梯度提升树GBDT系列算法

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
 
from sklearn.datasets import load_wine
from sklearn.ensemble import GradientBoostingClassifier as GBC
from sklearn.ensemble import RandomForestClassifier as RFC
from sklearn.tree import DecisionTreeClassifier as DTC
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score
 
X,y = load_wine(return_X_y=True,as_frame=True)
 
# 切分训练集和测试集
Xtrain, Xtest, Ytrain, Ytest = train_test_split(X,y,test_size=0.3,random_state=0)
 
# 使用GBDT完成对红酒数据集的预测
clf = GBC()    #实例化GBDT分类器，并使用默认参数
clf = clf.fit(Xtrain,Ytrain)
 
train_score = clf.score(Xtrain,Ytrain)
test_score = clf.score(Xtest,Ytest)
print(f"GBDT在训练集上的预测准确率为{train_score}")
print(f"GBDT在测试集上的预测准确率为{test_score}")

dtc = DTC(random_state=0) #实例化单棵决策树
dtc = dtc.fit(Xtrain,Ytrain)
score_dtc = dtc.score(Xtest,Ytest)
 
rfc = RFC(random_state=0) #实例化随机森林
rfc = rfc.fit(Xtrain,Ytrain)
score_rfc = rfc.score(Xtest,Ytest)
 
gbc = GBC(random_state=0) #实例化GBDT
gbc = gbc.fit(Xtrain,Ytrain)
score_gbc = gbc.score(Xtest,Ytest)
# 默认使用准确度（accuracy）作为评分方式,即预测正确的样本数占总样本数的比例
 
print("决策树：{}".format(score_dtc))
print("随机森林：{}".format(score_rfc))
print("GBDT：{}".format(score_gbc))

score_dtc = []
score_rfc = []
score_gbc = []
 
for i in range(10):
    dtc = DTC()
    cv1 = cross_val_score(dtc,X,y,cv=5)
    score_dtc.append(cv1.mean())
    
    rfc = RFC()
    cv2 = cross_val_score(rfc,X,y,cv=5)
    score_rfc.append(cv2.mean())
    
    gbc = GBC()
    cv3 = cross_val_score(gbc,X,y,cv=5)
    score_gbc.append(cv3.mean())
 
plt.plot(range(1,11),score_dtc,label = "DecisionTree")
plt.plot(range(1,11),score_rfc,label = "RandomForest")
plt.plot(range(1,11),score_gbc,label = "GBDT")
plt.legend(bbox_to_anchor=(1.4,1))
plt.show()

X,y = fetch_california_housing(return_X_y=True,as_frame=True)
 
Xtrain, Xtest, Ytrain, Ytest = train_test_split(X,y,test_size=0.3,random_state=0)
 
# 使用GBDT完成对加利福尼亚房屋数据集的预测
 
gbr = GBR(random_state=0) #实例化GBDT
gbr = gbr.fit(Xtrain,Ytrain)
r2_gbdt = gbr.score(Xtest,Ytest) # 回归器默认评估指标为R2
r2_gbdt
# 0.7826346388949185
 
# 计算GBDT回归器的评估指标：均方误差MSE
from sklearn.metrics import mean_squared_error
pred = gbr.predict(Xtest)
MSE = mean_squared_error(Ytest,pred)
MSE
 
# 0.28979949770874125

import time
modelname = ["DecisionTree","RandomForest","GBDT","RF-D"]
models = [DTR(random_state=0)
          ,RFR(random_state=0)
          ,GBR(random_state=0)
          ,RFR(random_state=0,max_depth=3)]
 
for name,model in zip(modelname,models):
    start = time.time()
    result = cross_val_score(model,X,y,cv=5,scoring="neg_mean_squared_error").mean()
    end = time.time()-start
    print(name)
    print("\t MSE:{:.3f}".format(abs(result)))
    print("\t time:{:.2f}s".format(end))
    print("\n")

DecisionTree
     MSE:0.818
     time:0.66s
 
 
RandomForest
     MSE:0.425
     time:70.69s
 
 
GBDT
     MSE:0.412
     time:16.84s
 
 
RF-D
     MSE:0.639
     time:11.49s